#aptible

Hey guys – getting a odd postgreSQL error when attemping to run a large data migration on our staging environment. Migration worked fine locally. Any of you guys have a second to take a look at it?

hey dleve123 — sure thing

(DMing)

Hey Aptible folks, trying out a deployment and getting the following: {"errorDetail":{"message":"The command [/bin/sh -c /build/builder] returned a non-zero code: 1"},"error":"The command [/bin/sh -c remote: uilder] returned a non-zero code: 1"}

Any insights?

hi iamryo!

that's Docker's output for when a command in your Dockerfile fails

Ahh...

if that's the exact output, it might be that somewhere "uilder" is being called instead of say "builder" ?

Yeah, weird. That's what I got back 'uilder'

Dockerfile stolen from a previous chat in here: FROM tutum/buildstep

PORT 3000

hmm — which app handle? i'll take a look

Sorry, disconnected there for a bit

it's iris-try-out

Build looks good locally too...

ah — spotted the likely culprit

seeing this in the logs (they should be in your git push output too, but maybe not?)

chfn: PAM: System error

adduser: `/usr/bin/chfn -f u27659' returned error code 1. Exiting.

My end or yours?

Ah, yup that's there

this happens on Linux 3.16, which we're running on your stack. the fix is the one recommended here : http://stackoverflow.com/questions/25193161/chf...

Do I need to do something with permissions?

Got it - I'll check that out.

Thanks

let me think through how you'd do that w/ tutum/buildstep

One of the answers suggests adding this to the Dockerfile: RUN ln -s -f /bin/true /usr/bin/chfn

Trying that now so

Nope, that didn't do it

yeah, give me a sec and i'll give you the Dockerfile you'll need

Cool, thanks

the problem is that tutum/buildstep runs its steps in ONBUILD, which will happen before anything you put in your Dockerfile

I can try another image too. The Passenger on should work, right?

not necessarily, at least not out of the box

got it

buildstep tries to determine your dependencies

ah, I didn't even look at it

i'd try this: https://gist.github.com/fancyremarker/612ecd850...

that should build — the only piece i'm not sure about is the ENTRYPOINT /exec

if that doesn't work, a solution would be to remove that from the Dockerfile and change the Procfile to run (e.g.) `/exec rails server`

Looks like that did it

Thanks for the help

np!

Hmm... Looks like it deployed, but the "remote: INFO -- : Waiting for health check on web..." never seemed to return anything so it failed

What's it check for? 200?

any HTTP response

Is there a way to tell what failed?

i think this is what i was mentioning re: the /exec ENTRYPOINT

i don't think that's quite right

(taking a look at buildstep...)

I see

So maybe try adding that to the procfile?

that would've been my suggestion, yes, but it seems that for your particular app, it's not properly detecting that it's a Rails app. i'm not sure why, to start

have you deployed it on Heroku?

Weird

Yeah

I have to run but I'll poke around

you have my email from the app id?

yes, will email you with any info i come up with

Awesome, thanks again!